---
title: "COVID-Messaging Dashboard"
date: "2020-07"
output: 
  flexdashboard::flex_dashboard:
    orientation: columns
    source_code: embed
    vertical_layout: fill
    theme: "cerulean"
---

```{r setup, include=FALSE}
library(plyr) 
library(leaflet)
library(leafpop)
library(purrr)
library(plotly)
library(sf)
library(cartogram)
library(scales)
library(sjmisc)
library(shiny)
# library(albersusa)
library(highcharter)
library(dplyr)
library(viridisLite)
library(forecast)
library(treemap)
library(flexdashboard)
library(kableExtra)

library(rtweet)        # Used for extracting the tweets
library(tm)            # Text mining cleaning
library(stringr)       # Removing characters
library(qdapRegex)     # Removing URLs 
library(wordcloud2) 
library(wordcloud)
library(ggwordcloud)
`%>%` <- dplyr::`%>%`

data("usgeojson")

# Load plotting data
# load("~/Dropbox (Yale_FES)/YPCCC_19XW/Crockett/Prototype/R_Dashboard/plot_data.rda")
load("plot_data.rda")

st.tweets <- tweet.counts[!duplicated(tweet.counts$state),]
names(st.tweets)[names(st.tweets)=='name'] <- 'name.governor'
st.tweets <- st.tweets %>% 
  filter(state %in% c("Alabama", "Alaska", "Arizona", "Arkansas", "California", "Colorado", 
"Connecticut", "Delaware", "Florida", "Georgia", "Hawaii", "Idaho", 
"Illinois", "Indiana", "Iowa", "Kansas", "Kentucky", "Louisiana", 
"Maine", "Maryland", "Massachusetts", "Michigan", "Minnesota", 
"Mississippi", "Missouri", "Montana", "Nebraska", "Nevada", "New Hampshire", 
"New Jersey", "New Mexico", "New York", "North Carolina", "North Dakota", 
"Ohio", "Oklahoma", "Oregon", "Pennsylvania", "Rhode Island", 
"South Carolina", "South Dakota", "Tennessee", "Texas", "Utah", 
"Vermont", "Virginia", "Washington", "West Virginia", "Wisconsin", 
"Wyoming")) %>% 
  select(state, party, name.governor, n)

```


Classifed Tweets Display
=======================================================================


Column {.tabset}
-----------------------------------------------------------------------

### Lockdown vs Economy

#### Deontological %

```{r}
# thm <- 
#   hc_theme(
#     chart = list(
#       backgroundColor = "transparent",
#       style = list(fontFamily = "Source Sans Pro")
#     ),
#     xAxis = list(
#       gridLineWidth = 1)
#     )

# Map tweets volumn
# highchart() %>%
#   hc_add_series_map(usgeojson, st.tweets, name = "State",
#                     value = "n", joinBy = c("woename", "state"),
#                     dataLabels = list(enabled = TRUE,
#                                       format = '{point.properties.postalcode}')) %>%
#   # hc_colorAxis(stops = colstops) %>%
#   hc_legend(valueDecimals = 0, valueSuffix = "%") %>%
#   hc_mapNavigation(enabled = TRUE) %>%
#   hc_add_theme(thm)


n <- 5
stops1 <- data.frame(q = 0:n/n,
                    c = brewer_pal(palette = "Greens")(6),
                    stringsAsFactors = FALSE)
util.stops <- list_parse2(stops1)

stops2 <- data.frame(q = 0:n/n,
                    c = brewer_pal(palette = "Purples")(6),
                    stringsAsFactors = FALSE)
deon.stops <- list_parse2(stops2)

s.plot <- s.state %>% 
  filter(variable=="lockdown vs. opening the economy")
  
highchart() %>%
  hc_add_series_map(usgeojson, s.plot, name = "Deontological %",
                    value = "deon.pct", joinBy = c("woename", "State_name"),
                    dataLabels = list(enabled = TRUE,
                                      format = '{point.properties.postalcode}')) %>%
  hc_colorAxis(minColor="#ffffff", maxColor="#6a51a3", stops = deon.stops, min=0, max=100) %>%
  hc_legend(valueDecimals = 0, valueSuffix = "%") %>%
  hc_mapNavigation(enabled = TRUE) %>%
  hc_add_theme(hc_theme_smpl())

```


#### Tweets/Statements 

```{r}
df.display <- df %>%
  select(date, source_person, State_name,  deon_text, format,  `lockdown vs. opening the economy`) %>%
  filter(!is.na(deon_text)) %>% 
  filter(`lockdown vs. opening the economy` == "D") %>% 
  select(-`lockdown vs. opening the economy`) %>% 
  rename('Date' = 'date', 'Person' = 'source_person',
         'State' = 'State_name', 'Source' = 'format',
         'Text' = 'deon_text')

 kable(df.display) %>%
  kable_styling() %>%
  scroll_box(width='550px')

```

### Medical Care

#### Deontological %

```{r}

s.plot <- s.state %>% 
  filter(variable=="rationing medical care")

highchart() %>%
  hc_add_series_map(usgeojson, s.plot, name = "Deontological %",
                    value = "deon.pct", joinBy = c("woename", "State_name"),
                    dataLabels = list(enabled = TRUE,
                                      format = '{point.properties.postalcode}')) %>%
  hc_colorAxis(minColor="#ffffff", maxColor="#6a51a3", stops = deon.stops, min=0, max=100) %>%
  hc_legend(valueDecimals = 0, valueSuffix = "%") %>%
  hc_mapNavigation(enabled = TRUE) %>%
  hc_add_theme(hc_theme_smpl())

```


#### Tweets/Statements

```{r}
df.display <- df %>%
  select(date, source_person, State_name,  deon_text, format,  `rationing medical care`) %>%
  filter(!is.na(deon_text)) %>% 
  filter(`rationing medical care` == "D") %>% 
  select(-`rationing medical care`) %>% 
  rename('Date' = 'date', 'Person' = 'source_person',
         'State' = 'State_name', 'Source' = 'format',
         'Text' = 'deon_text')

kable(df.display) %>%
  kable_styling() %>%
  scroll_box(width='550px')

```


### Contact Tracing

#### Deontological %

```{r}

s.plot <- s.state %>% 
  filter(variable=="contact tracing & surveillance")

highchart() %>%
  hc_add_series_map(usgeojson, s.plot, name = "Deontological %",
                    value = "deon.pct", joinBy = c("woename", "State_name"),
                    dataLabels = list(enabled = TRUE,
                                      format = '{point.properties.postalcode}')) %>%
  hc_colorAxis(minColor="#ffffff", maxColor="#6a51a3", stops = deon.stops, min=0, max=100) %>%
  hc_legend(valueDecimals = 0, valueSuffix = "%") %>%
  hc_mapNavigation(enabled = TRUE) %>%
  hc_add_theme(hc_theme_smpl())

```


#### Tweets/Statements

```{r}
df.display <- df %>%
  select(date, source_person, State_name,  deon_text, format,  `contact tracing & surveillance`) %>%
  filter(!is.na(deon_text)) %>% 
  filter(`contact tracing & surveillance` == "D") %>% 
  select(-`contact tracing & surveillance`) %>% 
  rename('Date' = 'date', 'Person' = 'source_person',
         'State' = 'State_name', 'Source' = 'format',
         'Text' = 'deon_text')

kable(df.display) %>%
  kable_styling() %>%
  scroll_box(width='550px')

```


### Vaccine Enforcement

#### Deontological %

```{r}

s.plot <- s.state %>% 
  filter(variable=="vaccine enforcement")

highchart() %>%
  hc_add_series_map(usgeojson, s.plot, name = "Deontological %",
                    value = "deon.pct", joinBy = c("woename", "State_name"),
                    dataLabels = list(enabled = TRUE,
                                      format = '{point.properties.postalcode}')) %>%
  hc_colorAxis(minColor="#ffffff", maxColor="#6a51a3", stops = deon.stops, min=0, max=100) %>%
  hc_legend(valueDecimals = 0, valueSuffix = "%") %>%
  hc_mapNavigation(enabled = TRUE) %>%
  hc_add_theme(hc_theme_smpl())

```


#### Tweets/Statements

```{r}
df.display <- df %>%
  select(date, source_person, State_name,  deon_text, format,  `vaccine enforcement`) %>%
  filter(!is.na(deon_text)) %>% 
  filter(`vaccine enforcement` == "D") %>% 
  select(-`vaccine enforcement`) %>% 
  rename('Date' = 'date', 'Person' = 'source_person',
         'State' = 'State_name', 'Source' = 'format',
         'Text' = 'deon_text')

kable(df.display) %>%
  kable_styling() %>%
  scroll_box(width='550px')

```


### PPE

#### Deontological %

```{r}

s.plot <- s.state %>% 
  filter(variable=="PPE/Healthcare allocation")

highchart() %>%
  hc_add_series_map(usgeojson, s.plot, name = "Deontological %",
                    value = "deon.pct", joinBy = c("woename", "State_name"),
                    dataLabels = list(enabled = TRUE,
                                      format = '{point.properties.postalcode}')) %>%
  hc_colorAxis(minColor="#ffffff", maxColor="#6a51a3", stops = deon.stops, min=0, max=100) %>%
  hc_legend(valueDecimals = 0, valueSuffix = "%") %>%
  hc_mapNavigation(enabled = TRUE) %>%
  hc_add_theme(hc_theme_smpl())

```


#### Tweets/Statements

```{r}
df.display <- df %>%
  select(date, source_person, State_name,  deon_text, format,  `PPE/Healthcare allocation`) %>%
  filter(!is.na(deon_text)) %>% 
  filter(`PPE/Healthcare allocation` == "D") %>% 
  select(-`PPE/Healthcare allocation`) %>% 
  rename('Date' = 'date', 'Person' = 'source_person',
         'State' = 'State_name', 'Source' = 'format',
         'Text' = 'deon_text')

kable(df.display) %>%
  kable_styling() %>%
  scroll_box(width='550px')

```

### Other

#### Deontological %

```{r}

s.plot <- s.state %>% 
  filter(variable=="Other")

highchart() %>%
  hc_add_series_map(usgeojson, s.plot, name = "Deontological %",
                    value = "deon.pct", joinBy = c("woename", "State_name"),
                    dataLabels = list(enabled = TRUE,
                                      format = '{point.properties.postalcode}')) %>%
  hc_colorAxis(minColor="#ffffff", maxColor="#6a51a3", stops = deon.stops, min=0, max=100) %>%
  hc_legend(valueDecimals = 0, valueSuffix = "%") %>%
  hc_mapNavigation(enabled = TRUE) %>%
  hc_add_theme(hc_theme_smpl())

```


#### Tweets/Statements

```{r}
df.display <- df %>%
  select(date, source_person, State_name,  deon_text, format,  `Other`) %>%
  filter(!is.na(deon_text)) %>% 
  filter(`Other` == "D") %>% 
  select(-`Other`) %>% 
  rename('Date' = 'date', 'Person' = 'source_person',
         'State' = 'State_name', 'Source' = 'format',
         'Text' = 'deon_text')

kable(df.display) %>%
  kable_styling() %>%
  scroll_box(width='550px')

```



Column {.tabset}
-----------------------------------------------------------------------

### Lockdown vs Economy

#### Consequentialist %

```{r}

s.plot <- s.state %>% 
  filter(variable=="lockdown vs. opening the economy")

highchart() %>%
  hc_add_series_map(usgeojson, s.plot, name = "Consequentialist %",
                    value = "util.pct", joinBy = c("woename", "State_name"),
                    dataLabels = list(enabled = TRUE,
                                      format = '{point.properties.postalcode}')) %>%
  hc_colorAxis(minColor="#ffffff", maxColor="#238b45", stops = util.stops, min=0, max=100) %>%
  hc_legend(valueDecimals = 0, valueSuffix = "%") %>%
  hc_mapNavigation(enabled = TRUE) %>%
  hc_add_theme(hc_theme_smpl())

```


#### Tweets/Statements

```{r}
df.display <- df %>%
  select(date, source_person, State_name,  util_text, format, `lockdown vs. opening the economy`) %>%
  filter(!is.na(util_text)) %>% 
  filter(`lockdown vs. opening the economy` == "U") %>% 
  select(-`lockdown vs. opening the economy`) %>% 
  rename('Date' = 'date', 'Person' = 'source_person',
         'State' = 'State_name', 'Source' = 'format',
         'Text' = 'util_text')

kable(df.display) %>%
  kable_styling() %>%
  scroll_box(width='550px')

```

### Medical Care

#### Consequentialist %

```{r}

s.plot <- s.state %>% 
  filter(variable=="rationing medical care")

highchart() %>%
  hc_add_series_map(usgeojson, s.plot, name = "Consequentialist %",
                    value = "util.pct", joinBy = c("woename", "State_name"),
                    dataLabels = list(enabled = TRUE,
                                      format = '{point.properties.postalcode}')) %>%
  hc_colorAxis(minColor="#ffffff", maxColor="#238b45", stops = util.stops, min=0, max=100) %>%
  hc_legend(valueDecimals = 0, valueSuffix = "%") %>%
  hc_mapNavigation(enabled = TRUE) %>%
  hc_add_theme(hc_theme_smpl())

```


#### Tweets/Statements

```{r}
df.display <- df %>%
  select(date, source_person, State_name,  util_text, format,  `rationing medical care`) %>%
  filter(!is.na(util_text)) %>% 
  filter(`rationing medical care` == "U") %>% 
  select(-`rationing medical care`) %>% 
  rename('Date' = 'date', 'Person' = 'source_person',
         'State' = 'State_name', 'Source' = 'format',
         'Text' = 'util_text')

kable(df.display) %>%
  kable_styling() %>%
  scroll_box(width='550px')

```


### Contact Tracing

#### Consequentialist %

```{r}

s.plot <- s.state %>% 
  filter(variable=="contact tracing & surveillance")

highchart() %>%
  hc_add_series_map(usgeojson, s.plot, name = "Consequentialist %",
                    value = "util.pct", joinBy = c("woename", "State_name"),
                    dataLabels = list(enabled = TRUE,
                                      format = '{point.properties.postalcode}')) %>%
  hc_colorAxis(minColor="#ffffff", maxColor="#238b45", stops = util.stops, min=0, max=100) %>%
  hc_legend(valueDecimals = 0, valueSuffix = "%") %>%
  hc_mapNavigation(enabled = TRUE) %>%
  hc_add_theme(hc_theme_smpl())

```


#### Tweets/Statements

```{r}
df.display <- df %>%
  select(date, source_person, State_name,  util_text, format,  `contact tracing & surveillance`) %>%
  filter(!is.na(util_text)) %>% 
  filter(`contact tracing & surveillance` == "U") %>% 
  select(-`contact tracing & surveillance`) %>% 
  rename('Date' = 'date', 'Person' = 'source_person',
         'State' = 'State_name', 'Source' = 'format',
         'Text' = 'util_text')

kable(df.display) %>%
  kable_styling() %>%
  scroll_box(width='550px')

```


### Vaccine Enforcement

#### Consequentialist %

```{r}

s.plot <- s.state %>% 
  filter(variable=="vaccine enforcement")

highchart() %>%
  hc_add_series_map(usgeojson, s.plot, name = "Consequentialist %",
                    value = "util.pct", joinBy = c("woename", "State_name"),
                    dataLabels = list(enabled = TRUE,
                                      format = '{point.properties.postalcode}')) %>%
  hc_colorAxis(minColor="#ffffff", maxColor="#238b45", stops = util.stops, min=0, max=100) %>%
  hc_legend(valueDecimals = 0, valueSuffix = "%") %>%
  hc_mapNavigation(enabled = TRUE) %>%
  hc_add_theme(hc_theme_smpl())

```


#### Tweets/Statements

```{r}
df.display <- df %>%
  select(date, source_person, State_name,  util_text, format,  `vaccine enforcement`) %>%
  filter(!is.na(util_text)) %>% 
  filter(`vaccine enforcement` == "U") %>% 
  select(-`vaccine enforcement`) %>% 
  rename('Date' = 'date', 'Person' = 'source_person',
         'State' = 'State_name', 'Source' = 'format',
         'Text' = 'util_text')

kable(df.display) %>%
  kable_styling() %>%
  scroll_box(width='550px')

```


### PPE

#### Consequentialist %

```{r}

s.plot <- s.state %>% 
  filter(variable=="PPE/Healthcare allocation")

highchart() %>%
  hc_add_series_map(usgeojson, s.plot, name = "Consequentialist %",
                    value = "util.pct", joinBy = c("woename", "State_name"),
                    dataLabels = list(enabled = TRUE,
                                      format = '{point.properties.postalcode}')) %>%
  hc_colorAxis(minColor="#ffffff", maxColor="#238b45", stops = util.stops, min=0, max=100) %>%
  hc_legend(valueDecimals = 0, valueSuffix = "%") %>%
  hc_mapNavigation(enabled = TRUE) %>%
  hc_add_theme(hc_theme_smpl())

```


#### Tweets/Statements

```{r}
df.display <- df %>%
  select(date, source_person, State_name,  util_text, format,  `PPE/Healthcare allocation`) %>%
  filter(!is.na(util_text)) %>% 
  filter(`PPE/Healthcare allocation` == "U") %>% 
  select(-`PPE/Healthcare allocation`) %>% 
  rename('Date' = 'date', 'Person' = 'source_person',
         'State' = 'State_name', 'Source' = 'format',
         'Text' = 'util_text')

kable(df.display) %>%
  kable_styling() %>%
  scroll_box(width='550px')

```

### Other

#### Consequentialist %

```{r}

s.plot <- s.state %>% 
  filter(variable=="Other")

highchart() %>%
  hc_add_series_map(usgeojson, s.plot, name = "Consequentialist %",
                    value = "util.pct", joinBy = c("woename", "State_name"),
                    dataLabels = list(enabled = TRUE,
                                      format = '{point.properties.postalcode}')) %>%
  hc_colorAxis(minColor="#ffffff", maxColor="#238b45", stops = util.stops, min=0, max=100) %>%
  hc_legend(valueDecimals = 0, valueSuffix = "%") %>%
  hc_mapNavigation(enabled = TRUE) %>%
  hc_add_theme(hc_theme_smpl())

```


#### Tweets/Statements

```{r}
df.display <- df %>%
  select(date, source_person, State_name,  util_text, format,  `Other`) %>%
  filter(!is.na(util_text)) %>% 
  filter(`Other` == "U") %>% 
  select(-`Other`) %>% 
  rename('Date' = 'date', 'Person' = 'source_person',
         'State' = 'State_name', 'Source' = 'format',
         'Text' = 'util_text')

kable(df.display) %>%
  kable_styling("striped") %>%
  scroll_box(width='550px')

```





```{r}

# https://juliasilge.com/blog/tidy-text-classification/
  
```


Wordcloud
=======================================================================

Column {.tabset}
-----------------------------------------------------------------------

### Lockdown vs Economy

#### Deontological 

```{r}
df.text1 <- df %>% 
  select(date, source_person, State_name,  deon_text, format, `lockdown vs. opening the economy`) %>%
  filter(!is.na(deon_text)) %>% 
  filter(`lockdown vs. opening the economy` == "D") 

df.text1$deon_text <- gsub("\\s*\\([^\\)]+\\)","",as.character(df.text1$deon_text))
df.text <- df.text1$deon_text
df.text <- str_c(df.text, collapse = "")

df.text <-
  df.text %>%
  str_remove("\\n") %>%                   # remove linebreaks
  rm_twitter_url() %>%                    # Remove URLS
  rm_url() %>%
  str_remove_all("#\\S+") %>%             # Remove any hashtags
  str_remove_all("@\\S+") %>%             # Remove any @ mentions
  removeWords(stopwords("english")) %>%   # Remove common words (a, the, it etc.)
  removeNumbers() %>%
  stripWhitespace() %>%
  removeWords(c("amp"))                   # Final cleanup of other small changes

textCorpus <-
  Corpus(VectorSource(df.text)) %>%
  TermDocumentMatrix() %>%
  as.matrix()

textCorpus <- sort(rowSums(textCorpus), decreasing=TRUE)
textCorpus <- data.frame(word = names(textCorpus), freq=textCorpus, row.names = NULL)

# wordcloud2(data = textCorpus, minRotation = 0, maxRotation = 0, ellipticity = 0.6, shape = "circle", color = rep_len(brewer.pal(9, "Purples"), nrow(textCorpus)))

textCorpus$color <- cut(textCorpus$freq, 5)
ggplot(textCorpus, aes(label=word, size=freq, color=color)) +
  geom_text_wordcloud_area(rm_outside = TRUE, eccentricity = 1, area_corr_power = 1) + 
  scale_size_area(max_size = 10) +
  scale_color_manual(values =c("#BCBDDC", "#9E9AC8", "#807DBA", 
                               "#6A51A3", "#54278F", "#3F007D")) +
  theme_minimal() 


```

### Medical Care

#### Deontological 

```{r}
df.text1 <- df %>% 
  select(date, source_person, State_name,  deon_text, format, `rationing medical care`) %>%
  filter(!is.na(deon_text)) %>% 
  filter(`rationing medical care` == "D") 

df.text1$deon_text <- gsub("\\s*\\([^\\)]+\\)","",as.character(df.text1$deon_text))
df.text <- df.text1$deon_text
df.text <- str_c(df.text, collapse = "")

df.text <-
  df.text %>%
  str_remove("\\n") %>%                   # remove linebreaks
  rm_twitter_url() %>%                    # Remove URLS
  rm_url() %>%
  str_remove_all("#\\S+") %>%             # Remove any hashtags
  str_remove_all("@\\S+") %>%             # Remove any @ mentions
  removeWords(stopwords("english")) %>%   # Remove common words (a, the, it etc.)
  removeNumbers() %>%
  stripWhitespace() %>%
  removeWords(c("amp"))                   # Final cleanup of other small changes

textCorpus <-
  Corpus(VectorSource(df.text)) %>%
  TermDocumentMatrix() %>%
  as.matrix()

textCorpus <- sort(rowSums(textCorpus), decreasing=TRUE)
textCorpus <- data.frame(word = names(textCorpus), freq=textCorpus, row.names = NULL)

textCorpus$color <- cut(textCorpus$freq, 5)
ggplot(textCorpus, aes(label=word, size=freq, color=color)) +
  geom_text_wordcloud_area(shape="circle", rm_outside = TRUE) + 
  scale_size_area(max_size = 10) +
  scale_color_manual(values =c("#BCBDDC", "#9E9AC8", "#807DBA", 
                               "#6A51A3", "#54278F", "#3F007D")) +
  theme_minimal() 

```


### Contact Tracing

#### Deontological 

```{r}
df.text1 <- df %>% 
  select(date, source_person, State_name,  deon_text, format, `contact tracing & surveillance`) %>%
  filter(!is.na(deon_text)) %>% 
  filter(`contact tracing & surveillance` == "D") 

df.text1$deon_text <- gsub("\\s*\\([^\\)]+\\)","",as.character(df.text1$deon_text))
df.text <- df.text1$deon_text
df.text <- str_c(df.text, collapse = "")

df.text <-
  df.text %>%
  str_remove("\\n") %>%                   # remove linebreaks
  rm_twitter_url() %>%                    # Remove URLS
  rm_url() %>%
  str_remove_all("#\\S+") %>%             # Remove any hashtags
  str_remove_all("@\\S+") %>%             # Remove any @ mentions
  removeWords(stopwords("english")) %>%   # Remove common words (a, the, it etc.)
  removeNumbers() %>%
  stripWhitespace() %>%
  removeWords(c("amp"))                   # Final cleanup of other small changes

textCorpus <-
  Corpus(VectorSource(df.text)) %>%
  TermDocumentMatrix() %>%
  as.matrix()

textCorpus <- sort(rowSums(textCorpus), decreasing=TRUE)
textCorpus <- data.frame(word = names(textCorpus), freq=textCorpus, row.names = NULL)

textCorpus$color <- cut(textCorpus$freq, 5)
ggplot(textCorpus, aes(label=word, size=freq, color=color)) +
  geom_text_wordcloud_area(shape="circle", rm_outside = TRUE) + 
  scale_size_area(max_size = 10) +
  scale_color_manual(values =c("#BCBDDC", "#9E9AC8", "#807DBA", 
                               "#6A51A3", "#54278F", "#3F007D")) +
  theme_minimal() 

```



### Vaccine Enforcement

#### Deontological 

```{r, eval=FALSE}
df.text1 <- df %>% 
  select(date, source_person, State_name,  deon_text, format, `vaccine enforcement`) %>%
  filter(!is.na(deon_text)) %>% 
  filter(`vaccine enforcement` == "D") 

df.text1$deon_text <- gsub("\\s*\\([^\\)]+\\)","",as.character(df.text1$deon_text))
df.text <- df.text1$deon_text
df.text <- str_c(df.text, collapse = "")

df.text <-
  df.text %>%
  str_remove("\\n") %>%                   # remove linebreaks
  rm_twitter_url() %>%                    # Remove URLS
  rm_url() %>%
  str_remove_all("#\\S+") %>%             # Remove any hashtags
  str_remove_all("@\\S+") %>%             # Remove any @ mentions
  removeWords(stopwords("english")) %>%   # Remove common words (a, the, it etc.)
  removeNumbers() %>%
  stripWhitespace() %>%
  removeWords(c("amp"))                   # Final cleanup of other small changes

textCorpus <-
  Corpus(VectorSource(df.text)) %>%
  TermDocumentMatrix() %>%
  as.matrix()

textCorpus <- sort(rowSums(textCorpus), decreasing=TRUE)
textCorpus <- data.frame(word = names(textCorpus), freq=textCorpus, row.names = NULL)

textCorpus$color <- cut(textCorpus$freq, 5)
ggplot(textCorpus, aes(label=word, size=freq, color=color)) +
  geom_text_wordcloud_area(shape="circle", rm_outside = TRUE) + 
  scale_size_area(max_size = 10) +
  scale_color_manual(values =c("#BCBDDC", "#9E9AC8", "#807DBA", 
                               "#6A51A3", "#54278F", "#3F007D")) +
  theme_minimal() 

```

### PPE

#### Deontological 

```{r}
df.text1 <- df %>% 
  select(date, source_person, State_name,  deon_text, format, `PPE/Healthcare allocation`) %>%
  filter(!is.na(deon_text)) %>% 
  filter(`PPE/Healthcare allocation` == "D") 

df.text1$deon_text <- gsub("\\s*\\([^\\)]+\\)","",as.character(df.text1$deon_text))
df.text <- df.text1$deon_text
df.text <- str_c(df.text, collapse = "")

df.text <-
  df.text %>%
  str_remove("\\n") %>%                   # remove linebreaks
  rm_twitter_url() %>%                    # Remove URLS
  rm_url() %>%
  str_remove_all("#\\S+") %>%             # Remove any hashtags
  str_remove_all("@\\S+") %>%             # Remove any @ mentions
  removeWords(stopwords("english")) %>%   # Remove common words (a, the, it etc.)
  removeNumbers() %>%
  stripWhitespace() %>%
  removeWords(c("amp"))                   # Final cleanup of other small changes

textCorpus <-
  Corpus(VectorSource(df.text)) %>%
  TermDocumentMatrix() %>%
  as.matrix()

textCorpus <- sort(rowSums(textCorpus), decreasing=TRUE)
textCorpus <- data.frame(word = names(textCorpus), freq=textCorpus, row.names = NULL)

textCorpus$color <- cut(textCorpus$freq, 5)
ggplot(textCorpus, aes(label=word, size=freq, color=color)) +
  geom_text_wordcloud_area(shape="circle", rm_outside = TRUE) + 
  scale_size_area(max_size = 10) +
  scale_color_manual(values =c("#BCBDDC", "#9E9AC8", "#807DBA", 
                               "#6A51A3", "#54278F", "#3F007D")) +
  theme_minimal() 

```

### Other

#### Deontological 

```{r}
df.text1 <- df %>% 
  select(date, source_person, State_name,  deon_text, format, `Other`) %>%
  filter(!is.na(deon_text)) %>% 
  filter(`Other` == "D") 

df.text1$deon_text <- gsub("\\s*\\([^\\)]+\\)","",as.character(df.text1$deon_text))
df.text <- df.text1$deon_text
df.text <- str_c(df.text, collapse = "")

df.text <-
  df.text %>%
  str_remove("\\n") %>%                   # remove linebreaks
  rm_twitter_url() %>%                    # Remove URLS
  rm_url() %>%
  str_remove_all("#\\S+") %>%             # Remove any hashtags
  str_remove_all("@\\S+") %>%             # Remove any @ mentions
  removeWords(stopwords("english")) %>%   # Remove common words (a, the, it etc.)
  removeNumbers() %>%
  stripWhitespace() %>%
  removeWords(c("amp"))                   # Final cleanup of other small changes

textCorpus <-
  Corpus(VectorSource(df.text)) %>%
  TermDocumentMatrix() %>%
  as.matrix()

textCorpus <- sort(rowSums(textCorpus), decreasing=TRUE)
textCorpus <- data.frame(word = names(textCorpus), freq=textCorpus, row.names = NULL)

textCorpus$color <- cut(textCorpus$freq, 5)
ggplot(textCorpus, aes(label=word, size=freq, color=color)) +
  geom_text_wordcloud_area(shape="circle", rm_outside = TRUE) + 
  scale_size_area(max_size = 10) +
  scale_color_manual(values =c("#BCBDDC", "#9E9AC8", "#807DBA", 
                               "#6A51A3", "#54278F", "#3F007D")) +
  theme_minimal() 

```


Column {.tabset}
-----------------------------------------------------------------------

### Lockdown vs Economy

#### Consequentialist 

```{r}
# devtools::install_github("gaospecial/wordcloud2")
# library(wordcloud2)

## Wordcloud not display - try ggwordcloud
# https://cran.r-project.org/web/packages/ggwordcloud/vignettes/ggwordcloud.html

df.text1 <- df %>% 
  select(date, source_person, State_name,  util_text, format, `lockdown vs. opening the economy`) %>%
  filter(!is.na(util_text)) %>% 
  filter(`lockdown vs. opening the economy` == "U") 

# remove parentheses and text within 
df.text1$util_text <- gsub("\\s*\\([^\\)]+\\)","",as.character(df.text1$util_text))

df.text <- df.text1$util_text
df.text <- str_c(df.text, collapse = "")

df.text <-
  df.text %>%
  str_remove("\\n") %>%                   # remove linebreaks
  rm_twitter_url() %>%                    # Remove URLS
  rm_url() %>%
  str_remove_all(" \\(.*\\)") %>% 
  str_remove_all("#\\S+") %>%             # Remove any hashtags
  str_remove_all("@\\S+") %>%             # Remove any @ mentions
  removeWords(stopwords("english")) %>%   # Remove common words (a, the, it etc.)
  removeNumbers() %>%
  stripWhitespace() %>%
  removeWords(c("amp"))                   # Final cleanup of other small changes

textCorpus <-
  Corpus(VectorSource(df.text)) %>%
  TermDocumentMatrix() %>%
  as.matrix()

textCorpus <- sort(rowSums(textCorpus), decreasing=TRUE)
textCorpus <- data.frame(word = names(textCorpus), freq=textCorpus, row.names = NULL)

# wordcloud2(data = textCorpus, minRotation = 0, maxRotation = 0, ellipticity = 0.6, shape = "circle", color = rep_len(rev(brewer.pal(9, "Greens")), nrow(textCorpus)))

# # screenshot the wordcloud and display as a png (not working)
# markdown_widget <- function(widget,path=getwd(),filename="file.png"){
#   require(htmlwidgets)
#   require(webshot)
#   saveWidget(widget,"tmp.html",selfcontained = F)
#   file <- paste(path,filename,sep = "/")
#   webshot("tmp.html",file,delay = 5,vwidth = 1024,vheight = 768)
#   file.remove("tmp.html")
#   paste0("![",file,"](",file,")\n\n") %>% cat()
# }
# 
# widget <- wordcloud2(data = textCorpus, minRotation = 0, maxRotation = 0, ellipticity = 0.6, shape = "circle", color = rep_len(rev(brewer.pal(9, "Greens")), nrow(textCorpus)))
# markdown_widget(widget,path=".",filename="wordcloud.png")

textCorpus$color <- cut(textCorpus$freq, 5)
ggplot(textCorpus, aes(label=word, size=freq, color=color)) +
  geom_text_wordcloud_area(shape="circle", rm_outside = TRUE) + 
  scale_size_area(max_size = 10) +
  scale_color_manual(values =c("#A1D99B", "#74C476", "#41AB5D", 
                               "#238B45", "#006D2C", "#00441B")) +
  theme_minimal() 



```



### Lockdown vs Economy

#### Consequentialist 

```{r}
df.text1 <- df %>% 
  select(date, source_person, State_name,  util_text, format, `lockdown vs. opening the economy`) %>%
  filter(!is.na(util_text)) %>% 
  filter(`lockdown vs. opening the economy` == "U") 

df.text1$util_text <- gsub("\\s*\\([^\\)]+\\)","",as.character(df.text1$util_text))
df.text <- df.text1$util_text
df.text <- str_c(df.text, collapse = "")

df.text <-
  df.text %>%
  str_remove("\\n") %>%                   # remove linebreaks
  rm_twitter_url() %>%                    # Remove URLS
  rm_url() %>%
  str_remove_all("#\\S+") %>%             # Remove any hashtags
  str_remove_all("@\\S+") %>%             # Remove any @ mentions
  removeWords(stopwords("english")) %>%   # Remove common words (a, the, it etc.)
  removeNumbers() %>%
  stripWhitespace() %>%
  removeWords(c("amp"))                   # Final cleanup of other small changes

textCorpus <-
  Corpus(VectorSource(df.text)) %>%
  TermDocumentMatrix() %>%
  as.matrix()

textCorpus <- sort(rowSums(textCorpus), decreasing=TRUE)
textCorpus <- data.frame(word = names(textCorpus), freq=textCorpus, row.names = NULL)

# wordcloud2(data = textCorpus, minRotation = 0, maxRotation = 0, ellipticity = 0.6, shape = "circle", color = rep_len(brewer.pal(9, "Purples"), nrow(textCorpus)))

textCorpus$color <- cut(textCorpus$freq, 5)
ggplot(textCorpus, aes(label=word, size=freq, color=color)) +
  geom_text_wordcloud_area(shape="circle", rm_outside = TRUE) + 
  scale_size_area(max_size = 10) +
  scale_color_manual(values =c("#A1D99B", "#74C476", "#41AB5D", 
                               "#238B45", "#006D2C", "#00441B")) +
  theme_minimal() 


```

### Medical Care

#### Consequentialist 

```{r}
df.text1 <- df %>% 
  select(date, source_person, State_name,  util_text, format, `rationing medical care`) %>%
  filter(!is.na(util_text)) %>% 
  filter(`rationing medical care` == "U") 

df.text1$util_text <- gsub("\\s*\\([^\\)]+\\)","",as.character(df.text1$util_text))
df.text <- df.text1$util_text
df.text <- str_c(df.text, collapse = "")

df.text <-
  df.text %>%
  str_remove("\\n") %>%                   # remove linebreaks
  rm_twitter_url() %>%                    # Remove URLS
  rm_url() %>%
  str_remove_all("#\\S+") %>%             # Remove any hashtags
  str_remove_all("@\\S+") %>%             # Remove any @ mentions
  removeWords(stopwords("english")) %>%   # Remove common words (a, the, it etc.)
  removeNumbers() %>%
  stripWhitespace() %>%
  removeWords(c("amp"))                   # Final cleanup of other small changes

textCorpus <-
  Corpus(VectorSource(df.text)) %>%
  TermDocumentMatrix() %>%
  as.matrix()

textCorpus <- sort(rowSums(textCorpus), decreasing=TRUE)
textCorpus <- data.frame(word = names(textCorpus), freq=textCorpus, row.names = NULL)

textCorpus$color <- cut(textCorpus$freq, 5)
ggplot(textCorpus, aes(label=word, size=freq, color=color)) +
  geom_text_wordcloud_area(shape="circle", rm_outside = TRUE) + 
  scale_size_area(max_size = 10) +
  scale_color_manual(values =c("#A1D99B", "#74C476", "#41AB5D", 
                               "#238B45", "#006D2C", "#00441B")) +
  theme_minimal() 

```


### Contact Tracing

#### Consequentialist 

```{r}
df.text1 <- df %>% 
  select(date, source_person, State_name,  util_text, format, `contact tracing & surveillance`) %>%
  filter(!is.na(util_text)) %>% 
  filter(`contact tracing & surveillance` == "U") 

df.text1$util_text <- gsub("\\s*\\([^\\)]+\\)","",as.character(df.text1$util_text))
df.text <- df.text1$util_text
df.text <- str_c(df.text, collapse = "")

df.text <-
  df.text %>%
  str_remove("\\n") %>%                   # remove linebreaks
  rm_twitter_url() %>%                    # Remove URLS
  rm_url() %>%
  str_remove_all("#\\S+") %>%             # Remove any hashtags
  str_remove_all("@\\S+") %>%             # Remove any @ mentions
  removeWords(stopwords("english")) %>%   # Remove common words (a, the, it etc.)
  removeNumbers() %>%
  stripWhitespace() %>%
  removeWords(c("amp"))                   # Final cleanup of other small changes

textCorpus <-
  Corpus(VectorSource(df.text)) %>%
  TermDocumentMatrix() %>%
  as.matrix()

textCorpus <- sort(rowSums(textCorpus), decreasing=TRUE)
textCorpus <- data.frame(word = names(textCorpus), freq=textCorpus, row.names = NULL)

textCorpus$color <- cut(textCorpus$freq, 5)
ggplot(textCorpus, aes(label=word, size=freq, color=color)) +
  geom_text_wordcloud_area(shape="circle", rm_outside = TRUE) + 
  scale_size_area(max_size = 10) +
  scale_color_manual(values =c("#A1D99B", "#74C476", "#41AB5D", 
                               "#238B45", "#006D2C", "#00441B")) +
  theme_minimal() 

```



### Vaccine Enforcement

#### Consequentialist 

```{r, eval=FALSE}
df.text1 <- df %>% 
  select(date, source_person, State_name,  util_text, format, `vaccine enforcement`) %>%
  filter(!is.na(util_text)) %>% 
  filter(`vaccine enforcement` == "U") 

df.text1$util_text <- gsub("\\s*\\([^\\)]+\\)","",as.character(df.text1$util_text))
df.text <- df.text1$util_text
df.text <- str_c(df.text, collapse = "")

df.text <-
  df.text %>%
  str_remove("\\n") %>%                   # remove linebreaks
  rm_twitter_url() %>%                    # Remove URLS
  rm_url() %>%
  str_remove_all("#\\S+") %>%             # Remove any hashtags
  str_remove_all("@\\S+") %>%             # Remove any @ mentions
  removeWords(stopwords("english")) %>%   # Remove common words (a, the, it etc.)
  removeNumbers() %>%
  stripWhitespace() %>%
  removeWords(c("amp"))                   # Final cleanup of other small changes

textCorpus <-
  Corpus(VectorSource(df.text)) %>%
  TermDocumentMatrix() %>%
  as.matrix()

textCorpus <- sort(rowSums(textCorpus), decreasing=TRUE)
textCorpus <- data.frame(word = names(textCorpus), freq=textCorpus, row.names = NULL)

textCorpus$color <- cut(textCorpus$freq, 5)
ggplot(textCorpus, aes(label=word, size=freq, color=color)) +
  geom_text_wordcloud_area(shape="circle", rm_outside = TRUE) + 
  scale_size_area(max_size = 10) +
  scale_color_manual(values =c("#A1D99B", "#74C476", "#41AB5D", 
                               "#238B45", "#006D2C", "#00441B")) +
  theme_minimal() 

```

### PPE

#### Consequentialist 

```{r, eval=FALSE}
df.text1 <- df %>% 
  select(date, source_person, State_name,  util_text, format, `PPE/Healthcare allocation`) %>%
  filter(!is.na(util_text)) %>% 
  filter(`PPE/Healthcare allocation` == "U") 

df.text1$util_text <- gsub("\\s*\\([^\\)]+\\)","",as.character(df.text1$util_text))
df.text <- df.text1$util_text
df.text <- str_c(df.text, collapse = "")

df.text <-
  df.text %>%
  str_remove("\\n") %>%                   # remove linebreaks
  rm_twitter_url() %>%                    # Remove URLS
  rm_url() %>%
  str_remove_all("#\\S+") %>%             # Remove any hashtags
  str_remove_all("@\\S+") %>%             # Remove any @ mentions
  removeWords(stopwords("english")) %>%   # Remove common words (a, the, it etc.)
  removeNumbers() %>%
  stripWhitespace() %>%
  removeWords(c("amp"))                   # Final cleanup of other small changes

textCorpus <-
  Corpus(VectorSource(df.text)) %>%
  TermDocumentMatrix() %>%
  as.matrix()

textCorpus <- sort(rowSums(textCorpus), decreasing=TRUE)
textCorpus <- data.frame(word = names(textCorpus), freq=textCorpus, row.names = NULL)

textCorpus$color <- cut(textCorpus$freq, 5)
ggplot(textCorpus, aes(label=word, size=freq, color=color)) +
  geom_text_wordcloud_area(shape="circle", rm_outside = TRUE) + 
  scale_size_area(max_size = 10) +
  scale_color_manual(values =c("#A1D99B", "#74C476", "#41AB5D", 
                               "#238B45", "#006D2C", "#00441B")) +
  theme_minimal() 

```

### Other

#### Consequentialist 

```{r}
df.text1 <- df %>% 
  select(date, source_person, State_name,  util_text, format, `Other`) %>%
  filter(!is.na(util_text)) %>% 
  filter(`Other` == "U") 

df.text1$util_text <- gsub("\\s*\\([^\\)]+\\)","",as.character(df.text1$util_text))
df.text <- df.text1$util_text
df.text <- str_c(df.text, collapse = "")

df.text <-
  df.text %>%
  str_remove("\\n") %>%                   # remove linebreaks
  rm_twitter_url() %>%                    # Remove URLS
  rm_url() %>%
  str_remove_all("#\\S+") %>%             # Remove any hashtags
  str_remove_all("@\\S+") %>%             # Remove any @ mentions
  removeWords(stopwords("english")) %>%   # Remove common words (a, the, it etc.)
  removeNumbers() %>%
  stripWhitespace() %>%
  removeWords(c("amp"))                   # Final cleanup of other small changes

textCorpus <-
  Corpus(VectorSource(df.text)) %>%
  TermDocumentMatrix() %>%
  as.matrix()

textCorpus <- sort(rowSums(textCorpus), decreasing=TRUE)
textCorpus <- data.frame(word = names(textCorpus), freq=textCorpus, row.names = NULL)

textCorpus$color <- cut(textCorpus$freq, 5)
ggplot(textCorpus, aes(label=word, size=freq, color=color)) +
  geom_text_wordcloud_area(shape="circle", rm_outside = TRUE) + 
  scale_size_area(max_size = 10) +
  scale_color_manual(values =c("#A1D99B", "#74C476", "#41AB5D", 
                               "#238B45", "#006D2C", "#00441B")) +
  theme_minimal() 

```



Lollipop
=======================================================================

```{r, fig.width=10, fig.height=8}
library(ggplot2)
library(dplyr)
library(hrbrthemes)


ggplot(s.state) +
  geom_segment( aes(x=variable, xend=variable, y=deon.pct, yend=util.pct), color="grey") +
  geom_point( aes(x=variable, y=deon.pct), color="#54278f", size=3 ) +
  geom_point( aes(x=variable, y=util.pct), color="#5aae61", size=3 ) +
  theme_ipsum() +
  facet_wrap(~State_name, nrow = 3)+
  coord_flip()+ 
  xlab("") +
  ylab("Purple - deontological % | Green - consequentialist %") +
  theme(legend.position = "none", axis.text = element_text(size=12), strip.text = element_text(size=15))


```